Is It Wrong for a Pastor to Use AI? A Working Pastor's Answer

No. Using AI is not wrong for a pastor, and I say that as a working priest who builds these tools for my own parish and teaches other pastors to do the same at Handbuilt.

The moral weight sits on two questions: which task you hand to the machine, and whether you tell the truth about having used it. Research, background reading, and first-draft support are ordinary uses of a tool. Generating a sermon and preaching it as your own unaided prayer is a different act with a different name.

The Short Answer

The line worth drawing runs between assistance and authorship. Assistance means the tool gathers and drafts while I stay the one who studies the text and stands up to say what it means for this congregation. That is a better reference shelf, and the work stays mine. Authorship means the machine composes the message and I preach it under my own name as though I had done the thinking, which misrepresents who did the work.

Almost every serious voice in this conversation sorts uses into those two buckets, even the ones who set the dial in very different places. John Dube at The Master's Seminary argues that sermon preparation is itself an act of worship and personal sanctification, so he keeps AI well clear of the study [1]. Ed Stetzer at Biola offers the image I reach for most: AI as the sous chef who preps and chops while the pastor stays the head chef who decides what the meal is [2]. John Piper allows AI as a source of information and forbids it as an undisclosed author, resting the whole question on honesty [3].

Pastors already feel this line before anyone explains it to them. Barna's research, reported by Religion News Service, found that only 12% of senior Protestant clergy are comfortable using AI to write sermons, while 43% see real value in it for preparation and research [4]. That gap is the instinct at work: comfort rises for the surrounding labor and drops as the task moves toward the pulpit.

How Many Pastors Are Already Doing This

This is not a fringe question asked by a handful of early adopters. Most pastors are already mid-experiment and quiet about it.

I hear this in ordinary conversation more than in any survey. Colleagues bring up a chat window the same way they used to bring up a commentary, a little sheepishly, as if admitting to a shortcut rather than describing a tool most of their peers are already reaching for.

The 2025 State of AI in the Church Survey Report, which polled 594 pastors and church staff, found 61% of pastors using AI weekly or daily, up from 43% the year before, with a quarter using it every day [5]. Most of that runs through general-purpose chat tools rather than anything built for ministry.

Tool	Share of pastors using it
ChatGPT	26%
Grammarly	11%
Microsoft Copilot	9%
Google Gemini	8%
Canva Magic Studio	8%

The same survey found 87% of pastors willing to invest in AI education or training, and separate Pushpay data pointed to an 80% jump in AI adoption across churches, even as leaders stayed cautious about using it for sermons and devotionals [5]. The appetite is already here. The open question is what disciplined use actually looks like.

Where the Positions Actually Fall

The public debate gets flattened into permission or prohibition, but the actual writing on this subject spreads across a range. Every serious voice I have read places AI somewhere between forbidden and unrestricted, disagreeing only about where the line sits.

Voice	Where they land	Core reasoning
John Dube, The Master's Seminary	Most restrictive	Sermon prep is worship and sanctification, and a general model cannot form a specific preacher for a specific congregation
Al Santos, Mid-America Baptist Seminary	AI-enhanced, not AI-dependent	A calculator analogy: useful as a tool, dangerous as a crutch that atrophies the skill
Ryan Hayden, working pastor	Moderate, rule-bound	Three rules: plagiarism is still plagiarism, the process matters, verify everything
Ed Stetzer, Biola University	Sous chef	AI supports the operation while the pastor stays head chef and answers for the result
Dave Betts, Church and AI	Assistant only	AI can speed the work but must never do the work: research and logic-checking, never the message
John Piper, Desiring God	Permissive for research	Information-gathering is acceptable, undisclosed composition is deception

Line those six up and the same boundary keeps appearing from a different angle each time. The tool is welcome around the edges of the work and suspect the moment it reaches for the center.

The Formation Problem

Several of these writers share a conviction that the struggle of preparation is not wasted time to be optimized away. The labor forms the preacher. Skip it and you may save an afternoon while losing something that never shows up on a clock.

The theological angle

Dube makes the sharpest version of the argument. The work of studying and analyzing Scripture is itself sanctifying, part of how God shapes the person who will preach, so handing that labor to a model shortcuts the formation it was meant to accomplish [1]. Ryan Hayden makes a gentler version, drawing on Haddon Robinson: the process of wrestling a text into a sermon changes the preacher, and that change is part of the point rather than an inefficiency to remove [6].

The cognitive-science angle

Al Santos, who directs IT at Mid-America Baptist Theological Seminary, raises a parallel worry from outside theology. He frames heavy reliance on generative tools as a risk of cognitive atrophy, pointing to research he reads as showing that people who offload their writing to a model retain little of what they produced [7]. I hold that specific research loosely, since I am passing along his reading of it rather than a study I checked myself. The underlying caution still lands for me. A skill you stop practicing is a skill you slowly lose.

Both angles land on the same practical rule I actually use. If a task builds me as a preacher, I do it myself and let the tool wait outside the study. If a task is genuinely administrative and has nothing to teach me, I hand it over and spend the reclaimed hour with people instead.

The Honesty Test

If formation is the argument I find most personally convincing, disclosure is the one nearly every source treats as non-negotiable. Piper puts it in the strongest terms. Presenting AI-generated content as your own creates a false impression in the minds of the people you serve, and he calls that deception plainly. He tested the risk on himself, having ChatGPT draft an 800-word response in his own theological style, and found it convincing enough that readers might not catch the difference, which he offered as a warning [3].

Hayden reaches the same rule through a different door. If it would be wrong to copy a passage from another person without credit, it is wrong to copy it from a model without credit [6]. The medium does not launder the borrowing.

Disclosure does not have to be a disclaimer read from the pulpit every Sunday. In practice it looks like a few ordinary habits:

• Being able to say, without flinching, exactly which parts of a piece came from a workflow and which came from you.
• Attributing the facts and quotes the tool surfaced, the same way you would credit a commentary.
• Never letting a generated paragraph stand in for study you did not do.
• Checking every fact and citation before it reaches anyone, because a model will produce a confident sentence whether or not it is true.
• Telling a colleague or a search committee the same thing you would tell your congregation, so the honesty does not stop at the pulpit.

What AI Cannot Know About Your People

Here is the red line almost no one crosses. A model does not know your people, and pastoral care depends entirely on knowing them.

Pastor Paul Hoffman put it in a question I have not been able to shake, quoted in the same Religion News Service reporting [4]:

Does AI know the stories of your people? Do they know about the miscarriage?

That is the whole thing. A general model can describe what grief support looks like in the abstract, but it has no way to know that the woman in the third pew stopped coming to the early service after her husband died, or why that detail matters more than any well-organized summary. Ed Stetzer names the same worry as a disconnect between the tool and a congregation it never met [2]. Dave Betts locates it in the work of interpretation and application, which depends on listening for the Spirit and knowing this specific body of people, and stays stubbornly human [8].

Sitting with a family in a hospital room, discerning in real time what they actually need: that work does not survive a handoff to a machine, and I would not want to serve a church that thought it should.

No tool has ever sat with a grieving family for me, and I do not expect one ever will. The parts of ministry that matter most are the parts a model cannot touch, which is exactly why I am relaxed about handing it the parts that were never sacred to begin with.

A Wider Lens: How Other Traditions Draw This Line

The debate in the search results is almost entirely a Protestant one, but the same instincts show up well outside it, which tells me the line is not a denominational quirk.

Rabbi Daniel Bogard, quoted in the same Religion News Service reporting behind that Barna finding, treats AI the way Jewish tradition treats a chavrutah, a study partner you argue with to sharpen your own understanding. The tool becomes a sparring partner for the text, and the learning still happens inside the student. That is close to how I actually use these tools in my own study.

Guidance from The Church of Jesus Christ of Latter-day Saints lands on the other recurring principle. The church says it will attribute AI-created content whenever its authenticity, accuracy, or authorship could otherwise be misunderstood [9]. Disclosure again, from a tradition with no stake in the evangelical Protestant argument.

The formation argument, the honesty argument, and the pastoral-care line all hold across traditions that agree on very little else. That consistency is part of why I trust them.

I did not go looking for that overlap to make a point. It showed up on its own while I was reading, and I hold the line with more confidence because it comes from people who share almost nothing else in common.

From Prompting to Workflows: Where I Actually Use It

Most pastors already using AI are stuck in the same place: a chat window. You type a question, read the answer, and start over next week from scratch, because the tool remembers nothing about how you work. That is real help, and it is also a ceiling.

The change that mattered in my ministry was moving from prompting to workflows, agents I set up once and then own, that run the repetitive parts of my week without me sitting there typing. Here is what that looks like in my parish, so you can see it is concrete:

• Citation-backed research and document drafting, turned into clean, finished PDFs.
• A full brand standards guide for the parish, plus business card and letterhead designs, built inside one week.
• Invoice-management automations that handle what used to cost me an afternoon.
• A maintained hymn database, with the fetching and organizing automated.
• Email responses drafted for my review, never sent unread.
• A bespoke dashboard that pulls from scattered systems and surfaces the real-time numbers I need to watch.

None of that touches a sermon, a hospital visit, or a pastoral conversation. All of it is the administrative and creative weight that used to sit between me and those things. This is the gap I built the Handbuilt Lab to close: teaching working pastors to build and own these workflows themselves, so they leave with something installed in their own ministry instead of one more subscription to depend on.

A Quick Self-Check Before You Use AI for Ministry Work

You do not need a committee to work this out. Before I use a tool for anything close to ministry, I run four questions, drawn from the guardrails Ryan Hayden and Ed Stetzer both land on.

Question	Why it matters
Would I tell my congregation exactly how I used this?	If the honest answer needs to stay hidden, that is already the answer
Am I still doing the reading and the wrestling myself?	The tool can gather, but it cannot do the study that forms the preacher
Does this bring me closer to my people or add distance?	Care and presence are the work, and they should never be automated away
Have I verified every fact and citation it gave me?	A model produces confident text with no regard for whether it is true

If a use clears all four, I do it and think no more about it. If it fails even one, I stop and figure out why.

That honest middle path is what I teach. If you are already using these tools and want to move from typing prompts to building workflows you own, without handing over your voice or your judgment, that is what the Handbuilt Lab is for. The first cohort begins the second week of September, capped at fifteen pastors, and runs by invitation rather than an open sign-up. Working out where your own line sits takes a real conversation about your ministry, so book a call and we can talk it through before anything else.

FAQ

Is it a sin for a pastor to use AI?

Using AI is not inherently sinful. The moral question turns on which task you delegate and whether you are honest about it. Using a model for research or first-draft support is an ordinary use of a tool. Generating a sermon and presenting it as if it were entirely your own unaided work is a question of honesty, and that is where the real caution belongs.

Can a pastor use AI to help write a sermon?

Most thoughtful voices in this debate allow AI for sermon research, background reading, and early drafting, while drawing a firm line at letting the model compose the sermon you deliver under your own name. John Piper permits information-gathering and forbids undisclosed composition. The safe practice is to let the tool assist your study while you remain the one who interprets the text and writes the message.

What should AI never be used for in ministry?

Pastoral care is the near-universal red line. A model does not know your congregation's specific griefs and needs, so counseling and presence with people in crisis stay human. As Pastor Paul Hoffman asks, AI does not know the stories of your people or which of them just lost a pregnancy.

Do most pastors already use AI?

Yes. The 2025 State of AI in the Church Survey found 61% of pastors using AI weekly or daily, up from 43% the year before, with ChatGPT the most common tool at 26% adoption. For most pastors the live question is how to use it with integrity, since they have already started.

Should a pastor tell the congregation they used AI?

Disclosure is the one rule almost every serious voice shares. You do not need a formal disclaimer every Sunday, but you should be able to say plainly how a tool contributed to your work, and you should attribute anything it surfaced the way you would credit a commentary. Ryan Hayden frames the standard simply: if copying from a person without credit would be wrong, copying from a model without credit is too.

Is using AI for sermon research different from using a commentary?

The comparison is closer than the alarm around it suggests. A commentary gives you one scholar's fixed reading of a passage; a model can synthesize across many sources and even draft language for you, which is why verification and disclosure matter more here, not less. Treat it as a faster reference shelf, never as a substitute for reading the text yourself.

References

[1] John Dube. "Don't Be an Artificial Preacher: An Argument Against Using AI in Sermon Preparation." The Master's Seminary Blog, January 23, 2025. https://blog.tms.edu/dont-be-an-artificial-preacher

[2] Ed Stetzer. "Navigating the Ethics of AI in Ministry and Sermon Writing." The Good Book Blog, Biola University, October 22, 2025. https://www.biola.edu/blogs/good-book-blog/2025/navigating-the-ethics-of-ai-in-ministry-and-sermon-writing

[3] John Piper. "Should I Use AI to Help Me Write Sermons?" Desiring God, February 24, 2025. https://www.desiringgod.org/interviews/should-i-use-ai-to-help-me-write-sermons

[4] Deena Prichep. "Are AI Sermons Ethical? Clergy Consider Where to Draw the Line." The Presbyterian Outlook (Religion News Service), August 4, 2025. https://pres-outlook.org/2025/08/are-ai-sermons-ethical-clergy-consider-where-to-draw-the-line/

[5] Leonardo Blair. "Majority of pastors now using AI to prepare sermons amid rapid embrace of technology: study." The Christian Post, December 7, 2025. https://www.christianpost.com/news/majority-of-pastors-using-ai-to-prepare-sermons-study.html

[6] Ryan Hayden. "Sermon Preparation in Seconds: Ethical Guidelines for Using AI in Sermon Preparation." HEY World, August 17, 2024. https://world.hey.com/ryan.hayden/sermon-preparation-in-seconds-ethical-guidelines-for-using-ai-in-sermon-preparation-4c70399c

[7] Al Santos. "Is It Okay to Use AI in Ministry?" Mid-America Baptist Theological Seminary, January 22, 2026. https://www.mabts.edu/2026/01/22/is-it-okay-to-use-ai-in-ministry/

[8] Dave Betts. "Can AI and Preaching Coexist? Three Ethical Ways to Use It Well." Church and AI, March 19, 2025. https://www.churchandai.com/p/can-ai-and-preaching-coexist-three

[9] "Guiding Principles for the Church's Use of Artificial Intelligence." Newsroom, The Church of Jesus Christ of Latter-day Saints. Accessed July 11, 2026. https://newsroom.churchofjesuschrist.org/article/church-jesus-christ-artificial-intelligence